3.1.6.1 Thyroid imaging: nonisotopic techniques

Clinical examination and evaluation of thyroid function remain fundamental in the evaluation of thyroid disorders, but observer variation leads to a considerable heterogeneity in the evaluation of patients with suspected thyroid disease (1). It is not surprising, therefore, that imaging of the thyroid is often performed. Although it most often cannot distinguish between benign and malignant lesions, and its clinical value is generally thought to be limited (2), a European survey demonstrated that 88% of European Thyroid Association members would use imaging in an index case of a euthyroid patient with a solitary thyroid nodule and absence of clinical suspicion of malignancy (3). In the case of a clinically benign nontoxic multinodular goitre, the figure was 91% (4).

The thyroid gland can be evaluated by several different nonisotopic imaging techniques. The most commonly used are ultrasonography, CT, and MRI. Each method has advantages and limitations, and there is no absolute clinical indication for performing any of these imaging procedures in the majority of patients. The major drawback of all techniques, in addition to expense, is that the technical advances in thyroid imaging have not been accompanied by increased specificity for tissue diagnosis. This chapter will focus on the clinical use of these methods and, as far as this is possible, compare their advantages and disadvantages (Table 3.1.6.1.1).

Examination of the neck is performed with high-frequency transducers (7–15 MHz), and the patient is in the supine position with the neck hyperextended. The transducer is coupled to the skin with gel since ultrasound does not pass through air. The technique can detect thyroid lesions as small as 2 mm. It can distinguish solid from simple and complex cysts. It enables the accurate determination of thyroid size, gives a rough estimate of echogenicity, visualizes vascular flow and velocity (colour flow Doppler), and aids in the accurate placing of needles, be it for diagnostic or therapeutic purposes (Box 3.1.6.1.1). The main drawbacks are the high degree of observer dependency and the inability to visualize retroclavicular or intrathoracic extension of the thyroid (2, 5). The average investigation rarely takes more than 10 min.

Ultrasonography is based on the emission of high-frequency sound waves and subsequent reflection as they pass through the tissue. The amplitude of the reflections of the sound waves is due to differences in the acoustic impedance of the various body tissues. The depth of tissue penetration is the least for high-frequency waves. Conversely, structural resolution is best. The frequency used to visualize the thyroid (7–15 MHz) is a compromise between the need for depth of penetration and that for resolution. The use of real-time allows the differentiation of static structures (thyroid, neck muscles, lymph nodes) from that of moving or pulsating structures (blood vessels, oesophagus) (2, 5).

The normal thyroid parenchyma has a characteristic homogeneous medium-level echogenicity (Fig. 3.1.6.1.1). The surrounding muscles have a lower echogenicity. Posterolaterally the thyroid is bordered by the sonolucent common carotid artery and internal jugular vein, and medially by the trachea. The oesophagus with its echogenic mucosa can be seen behind and to the left of the trachea.

A high proportion of people with a normal thyroid gland have small (1–3 mm) cystic or solid lesions, the frequency being higher in women, increasing with age, and varying between countries (5, 6). The importance of these abnormalities is unclear, but since incidental sonographic nodules (‘incidentalomas’) are very common, whereas thyroid cancer is not, a conservative/expectant approach is generally recommended. An incidentally disclosed nodule or cyst less than 1 cm in diameter in an asymptomatic individual with a normal neck palpation should generally not lead to biopsy or further investigations (5, 6).

Goitre, i.e. an enlarged thyroid gland, remains a clinical diagnosis. But this evaluation carries an inaccuracy of approximately 40% and cannot reliably be used for size determination (1). For this, two principally different methods are available. One employs the model of a rotation ellipsoid and can be modified to length × width × thickness × π/6 for each lobe, and carries an inaccuracy of 15–20% that increases with size and degree of irregularity (2,5). The other method is based on obtaining cross-sections of the entire thyroid gland. This method carries an inaccuracy of 5–10% and is less influenced by size and degree of irregularity (2, 5). The normal thyroid size (5–30 ml in adults) is positively related to weight and age, increases with decreasing iodine intake, and is influenced by a number of physiological as well as environmental factors (2, 5). Ultrasonography is the most sensitive technique for screening populations for goitre and is widely used for field studies (2, 5, 7).

Nonautoimmune nontoxic diffuse goitre appears diffusely enlarged with a uniform or discretely irregular echo pattern without nodules. Various degrees of hypoechogenicity may be evident, but, when marked, suggest the presence of autoimmunity. In Hashimoto’s thyroiditis, hypoechogenicity is always marked but may be inhomogeneous. Ultrasonography cannot differentiate between benign autoimmune thyroiditis and lymphoma or carcinoma. Therefore, goitre growth, especially in the l-thyroxine-treated patient, should raise suspicion of lymphoma and lead to biopsy or operation. In Graves’ disease, the thyroid is most often enlarged and the echo pattern homogeneous. Echogenicity can be normal to markedly decreased and the latter suggests a decreased probability of achieving remission on antithyroid drugs. Colour flow Doppler can demonstrate the rich vascularity and increased flow related to the degree of hyperthyroidism. Subacute thyroiditis leads to thyroid enlargement and areas of hypoechogenicity probably related to areas that are affected. Remission leads to normalization of size, but areas of hypoechogenicity may remain long after remission is obtained.

Multinodular goitres are often larger than diffuse goitres and a significant number (10–20%) have a substernal or intrathoracic extension which cannot be visualized since the bony thorax prevents penetration of sound waves. The echographic structure may be heterogeneous without well-defined nodules or composed of multiple nodules interspersed throughout a normal-appearing gland. Often areas of haemorrhage, necrosis, and calcifications are seen. Most patients evaluated for a single nodule have additional small thyroid nodules when examined by ultrasonography (5, 6). The echogenicity of the nodules may vary from hyper- to iso-, to hypoechoic, even within the same patient. The presence of multiple nodules identified by ultrasound examination (or any other imaging modality) does not exclude malignancy, it is just as likely as in the solitary nodule (6, 8). Therefore, especially in view of the increasing use of nonsurgical treatment for this disorder, (6, 8) fine-needle aspiration biopsy (FNAB) should be used liberally especially in the patient with a dominant or growing nodule. Ultrasound guidance is also recommended for selection of the most suspicious nodules (6).

Thyroid cysts are well-defined areas with greatly reduced or absent echogenicity. Varying degrees of echogenicity can often be seen due to debris or necrotic tissue. True simple cysts are extremely rare (approximately 1% of all nodules) and virtually always benign. Most, however, are complex cysts and are as likely to be a carcinoma as is a solid nodule (Fig. 3.1.6.1.2). Cystic degeneration is present in 20–30% of thyroid carcinomas and benign solid nodules. After ultrasound-guided aspiration a residual nodule should be biopsied. In case of benign cytology and recurrence of the cyst (which is seen in approximately 50% of the patients) ultrasound-guided treatment can be offered. Flushing with ethanol decreases recurrence rate (9). Malignancy cannot be excluded either by cytology of the cystic component or by the colour of the cyst fluid.

There are no specific characteristics that can differentiate benign thyroid nodules from thyroid carcinomas. Neither size, echogenicity, elasticity, the finding of a sonographic halo, calcifications, nor vascularization can with acceptable specificity be used for this purpose (2, 5, 6). Therefore, the most cost-effective investigation of these patients is fine-needle aspiration biopsy. In Europe most thyroidologists will use ultrasound-guided fine-needle aspiration biopsy (3) and this increases the likelihood of obtaining a sufficient sample.

No sonographic finding is characteristic of any type of thyroid carcinoma, and ultrasonography cannot differentiate benign from malignant nodules. Extrathyroidal extension of the tumour or lymphadenopathy may suggest thyroid carcinoma but it is not proof. The ultrasound appearance of thyroid carcinoma is highly variable. Generally, it is hypoechoic relative to normal thyroid and microcalcifications are often present. Since very small nodules of 2–3 mm can be detected, ultrasound examination is increasingly used in the follow-up of patients treated for thyroid carcinoma or at risk because of previous irradiation (e.g. post-Chernobyl). Characteristic sonolucent masses in the thyroid bed or adjacent tissues often suggest recurrent disease before this is clinically evident.

CT offers excellent anatomical resolution by increasing the distinction of differences in density between soft tissues. Density differences as small as 0.5% can be detected compared to the 5–10% of conventional radiographic techniques. The accurate measurement of the absorption of X-rays by tissues (attenuation) enables individual tissues to be studied (2).

The technique is highly sensitive but just as nonspecific as ultrasonography in differentiating benign from malignant disease. It can distinguish solid from simple and complex cysts and enables the accurate determination of thyroid size. It is superior to ultrasonography when examining retroclavicular/intrathoracic goitre and it is not as observer dependent. The drawbacks are cost, limited availability for this purpose, length of the investigation, cooperability (claustrophobia), and exposure to ionizing irradiation (1–4 rads; 0.01–0.04 Gy). The image is not dynamic and although possible, CT-guided biopsy is more cumbersome than with ultrasonography. Intravenous contrast media, to visualize vascular relationships, pose a risk of allergic reactions (Table 3.1.6.1.1).

CT depends on the attenuation of an X-ray beam as it passes through tissues. The extent of attenuation depends on the tissue constituents, and the brightness of each portion (pixel) of the final image is proportional to the degree that it attenuates the X-rays passing through it. The image is usually depicted in shades of grey. Density values are expressed in CT numbers (Hounsfield units, HU), which are related to the attenuation value of water. The high endogenous iodine content of the thyroid enables its visualization. The CT density of the thyroid is closely correlated with its iodine content and can be used to estimate it.

The normal thyroid gland is easily visualized on CT and its density is always higher than surrounding tissues. Differences in density reported from various countries reflect differences in iodine intake. There is no sex difference in density but it decreases with age and as a consequence of l-thyroxine treatment.

Disease in the thyroid usually leads to decreased ability to concentrate iodine, therefore, reduced density on CT is the hallmark of thyroid disease. The exact density measurements have not proved useful in distinguishing between various thyroid disorders. Thus, the CT image may be compatible with a certain diagnosis but rarely specific for it.

Nonautoimmune nontoxic diffuse goitre appears to be homogeneously enlarged with various degrees of hypodensity. Graves’ disease is characterized by a 50–70% decrease in density and may be slightly inhomogeneous. Hashimoto’s thyroiditis typically demonstrates an inhomogeneous iodine distribution and a 50% decrease in CT density which is lowest in hypothyroid individuals. Increasing goitre size is characteristically associated with decreasing density. Asymmetrical hypodense areas should raise the suspicion of lymphoma or carcinoma.

Subacute thyroiditis is also characterized by hypodensity and is focal or diffuse depending on the extent of the disease. In the initial phases, acute suppurative thyroiditis has no characteristic CT image; however, as infection progresses, loculated abscesses with hypodensity may appear.

Multinodular goitre is often an enlarged asymmetrical gland with multiple low density areas of varying degrees of discreteness. CT density is decreased but in an inhomogeneous way. After intravenous contrast, enhancement is obtained except for areas containing haemorrhage, necrosis, or cysts. Calcifications are seen in up to 50% of goitres. Compression of the trachea, oesophagus, and great vessels is easily ascertained and CT has found use especially in patients with monstrous and partly intrathoracic goitre, where it is ideal for the estimation of tracheal compression and quantitation of the intrathoracic extension of the goitre. Anatomical continuity with the cervical thyroid as well as a CT density greater than muscle, provides evidence of its thyroidal origin. Mediastinal lymphoma, lymphadenopathy, or thymus usually have markedly lower CT densities.

Simple cysts are hypodense lesions, smooth-walled, and surrounded by normal thyroid tissue. The density of cyst fluid is always less than muscle and contrast injection does not lead to enhancement. Complex cysts are easily distinguished from simple cysts.

Thyroid nodules are common, usually round or oval lesions of low density, and, as with ultrasonography, no CT characteristics will separate benign from malignant lesions (2). Invasive growth into surrounding structures and metastases to cervical lymph nodes are suggestive of carcinoma. Papillary and follicular carcinomas are usually irregular low-density lesions and calcifications are present in the majority. There may be slight enhancement after contrast injection. The CT feature of medullary thyroid carcinoma is a single or multiple low-density lesions of variable size in one or both lobes. Lesions of 1–2 mm in size can be detected. Calcification is less often seen than in papillary carcinoma. Patients with C-cell hyperplasia have normal CT scans.

Large irregular masses of low attenuation with central cystic or necrotic areas are suggestive of anaplastic carcinoma especially if calcification is pronounced. Again, these features may also be seen in benign multinodular goitre. Invasion of the trachea, cricoid, or thyroid cartilage, and growth into the tracheal lumen is highly suggestive of carcinoma. Both Hashimoto’s thyroiditis and thyroid lymphoma appear as masses of reduced density with little enhancement after contrast injection, and CT alone cannot make a distinction between them.

CT is of value in the follow-up of patients with thyroid cancer. Recurrence is evident as discrete low-density lesions within or outside the thyroid bed. Lymph node metastases typically have a regular rim, a core of central lucency, and no enhancement after intravenous contrast. CT is highly sensitive in detecting extrathyroidal spread of disease and therefore complementary to whole-body scanning with radioactive iodine.

Combined CT and positron emission tomography (PET) with [¹⁸F]2-fluoro-2-deoxy-d-glucose (FDG) is a novel multimodality technology that enables a more precise anatomical localization of an area with increased focal uptake (potentially malignant lesion). Its role in the initial evaluation of a thyroid nodule is limited, but can be of value in case of indeterminate cytology (10). It is increasingly used where there is suspicion of recurrence or spread of thyroid cancer (11).

MRI offers excellent anatomical resolution and generation of images in multiple planes. The technique is highly sensitive but just as nonspecific as ultrasonography and CT in differentiating benign from malignant lesions (2). It can distinguish solid from simple and complex cysts. It allows thyroid size determination and, as with CT and in contrast to ultrasonography, it can visualize the retrotracheal area and retroclavicular or intrathoracic goitre (12). Additionally, it is less operator dependent. The paramagnetic contrast agent gadolinium allows visualization of tumour vascularity. The drawbacks are cost, very limited availability for this purpose, length of the investigation, and cooperability (5% of patients cannot cooperate due to claustrophobia and some, especially children, need to be sedated). Patient and tissue movement (e.g. swallowing) decreases image quality and calcifications are better visualized with CT. MRI cannot be used in patients with cardiac pacemakers, implantable defibrillators, central nervous system aneurysmal clips, cochlear implants, and ferromagnetic ocular fragments. Small metal objects and orthopaedic devices decrease resolution and cause field inhomogeneity (Table 3.1.6.1.1).

MRI images depend on the magnetic properties of certain atomic nuclei. The MRI signal contains several variable components. T₁ relaxation time (longitudinal or spin-lattice relaxation time) reflects the time for protons to give up their energy to the surrounding environment (lattice) and return to their original alignment parallel to the magnetic field. The T₂ relaxation time (transverse or spin-spin relaxation time) is the time needed for synchronous transverse spinning to decay after excitation. Adjustment of the pulse sequence can favour one or the other of these magnetic properties.

On T₁-weighted images the normal thyroid gland is clearly seen on MRI and shows a nearly homogeneous signal with an intensity similar to that of the adjacent neck muscles. On T₂-weighted images, the normal thyroid has a much greater signal intensity than adjacent muscles. Blood vessels, lymph nodes, fat, and muscle are clearly identified and distinguished from the thyroid.

In Graves’ disease the thyroid has slightly heterogeneous diffusely increased signal on both T₁- and T₂-weighted images. Hashimoto’s thyroiditis causes a heterogeneous signal intensity on T₁-weighted images and a diffusely increased signal on T₂-weighted images.

MRI can detect nodules as small as 3–5 mm (Fig. 3.1.6.1.3). Characteristically multinodular goitres have various degrees of heterogeneity and low to increased signal intensity on T₁-weighted images. T₂-weighted images show more pronounced heterogeneity and increased intensity. Nodules are better visualized on T₂-weighted images.

Simple cysts have a low-intensity signal on both T₁- and T₂-weighted imaging. The intensity on T₁-weighted images increases with increasing protein and lipid content.

Follicular adenomas appear round or oval with a heterogeneous signal equal to or greater than that of normal tissue (2). On T₂-weighted images the nodules have increased signal intensity. No MRI characteristics will accurately separate benign from malignant lesions. Thyroid carcinomas appear as focal or nonfocal lesions of variable size; they are isointense or slightly hyperintense on T₁-weighted images and hyperintense on T₂-weighted images. The imaging characteristics of all types of thyroid carcinomas, including medullary carcinoma and lymphoma, are similar.

The extent of thyroid carcinoma can be determined preoperatively and may be useful in the planning of surgery. Extension into adjacent structures is usually evident. MRI cannot distinguish metastatic from inflammatory adenopathy, and both appear hyperintense on T₂-weighted images. Gadolinium may be useful since metastatic nodes are enhanced centrally after gadolinium injection. Furthermore, in the postoperative follow-up recurrent carcinoma enhance with gadolinium, whereas scarring generally does not.

Although there is no absolute clinical indication for performing any of the imaging procedures and although none of them can accurately distinguish benign from malignant disease they are increasingly used (2–5). Thyroid ultrasonography is the most commonly used technique. This is explained partly by increased availability and reduced cost, and also because a growing number of endocrinologists and internists, including the authors, have found it of value in several outpatient situations. Even if ultrasonography cannot reliably diagnose or exclude malignancy, it is of value in providing superior morphological detail compared to scintigraphy and in allowing the accurate placing of needles for diagnostic and therapeutic purposes. Additionally, it provides an accurate size determination and evaluation of echogenicity and thereby aids in the classification and follow-up of various thyroid disorders. As more patients with thyroid nodular disease are offered nonsurgical treatment, mainly in the form of radioiodine and ultrasound-guided percutaneous therapy with ethanol or laser (6, 8, 9), thyroid ultrasonography should become an integral part of the evaluation of many thyroid patients in the outpatient clinics of endocrinological departments (2, 5, 6).

CT and MRI can provide much of the information that is obtained with ultrasonography (2). Their greater expense, limited availability, and other drawbacks argue against their use most of the time. CT is valuable in determining the extent of a substernal goitre or in the evaluation of a mediastinal mass. It can give valuable information in the evaluation of thyroid carcinoma and its spread. MRI may be useful in the same setting and is generally superior to CT in the evaluation of recurrent carcinoma, be it in the thyroid bed or in regional lymph nodes (2).

Recently, ultrasound elastography, which uses ultrasound to provide an estimation of tissue stiffness by measuring the degree of distortion applied with the transducer, was introduced. Malignant nodules are more firm, and elastography is currently being evaluated as an adjunctive tool for the preoperative selection of thyroid nodules (13). In the follow-up of thyroid cancer, especially in high-risk patients, aside from the use of ultrasonography for the detection of local recurrence and cervical lymph node metastases, radioiodine imaging and FDG PET/CT are the methods of choice for localizing metastatic disease (10, 11).